A Deep Learning-Based Method and System for Oil Product Data Analysis

Abstract

This invention belongs to the field of data analysis technology and discloses a method and system for oil product data analysis based on deep learning. The method includes the following steps: using a full lifecycle data acquisition device to collect real-time multi-source oil product data from oil wells; using an edge computing gateway to preprocess the real-time multi-source oil product data and upload it to a cloud data center; in the cloud data center, using a multi-dimensional feature extraction model based on a dynamic multi-dimensional feature space to extract real-time merged multi-dimensional features; using an oil product data analysis model to analyze the real-time merged multi-dimensional features; and using the real-time oil product data analysis results to generate real-time production, refining, storage, and transportation strategies using a production, refining, storage, and transportation strategy generation model. This invention solves the problems of insufficient data processing capabilities, limited model depth and accuracy, and lack of response mechanisms in existing technologies.

Description

Technical Field

[0001] This invention belongs to the field of data analysis technology, specifically relating to a method and system for oil data analysis based on deep learning. Background Technology

[0002] Oil product data refers to the collection of various information related to oil and its products throughout the entire oil industry chain (from exploration, extraction, refining, storage, transportation to sales and use). This data is characterized by its multi-source heterogeneity, with wide sources, diverse formats, varying update frequencies, and complex physical, economic, and spatiotemporal relationships among the data. Effective analysis and utilization of this data are crucial for optimizing oil production planning, improving refining efficiency, reducing storage and transportation costs, ensuring supply chain stability, and enhancing market responsiveness. With the development of IoT, big data, and artificial intelligence technologies, leveraging advanced data analytics to improve the intelligence level of the oil industry has become an urgent need.

[0003] However, existing technologies still have many shortcomings, including:

[0004] 1) Insufficient data processing capabilities: Traditional data analysis methods, such as simple statistical analysis, regression models, or rule-based systems, can often only process data from a single source or a few closely related data, making it difficult to effectively integrate multi-source heterogeneous data from different stages such as mining, refining, storage, and transportation; some preliminary machine learning applications may perform feature selection, but they are usually based on a single objective or domain expert experience, lacking systematicity, automation, and intelligence.

[0005] 2) Limited model depth and accuracy: Existing technologies use simple models, such as support vector machines, traditional neural networks, or ensemble learning, to predict oil production, quality, or demand. When dealing with high-dimensional, nonlinear, and strongly correlated oil data, these models often struggle to capture complex patterns in the data. They also have limited ability to handle the dynamics and long-term dependencies of time series data, resulting in low prediction accuracy and failing to meet the needs of real-time decision-making.

[0006] 3) Lack of response mechanism: Existing technologies only realize a single data analysis function and lack an optimization response mechanism for the entire life cycle, resulting in low value of data analysis results and an inability to dynamically adjust production, storage and transportation processes based on data analysis results. Summary of the Invention

[0007] To address the problems of insufficient data processing capabilities, limited model depth and accuracy, and lack of response mechanisms in existing technologies, the present invention aims to provide a deep learning-based method and system for oil product data analysis.

[0008] The technical solution adopted in this invention is as follows:

[0009] A deep learning-based method for oil product data analysis includes the following steps:

[0010] Using a full lifecycle data acquisition device, real-time multi-source oil data from oil wells is collected and transmitted to an edge computing gateway within the communication range.

[0011] Using an edge computing gateway, real-time multi-source oil data is preprocessed, and the preprocessed real-time multi-source oil data is uploaded to the cloud data center.

[0012] In the cloud data center, based on the dynamic multidimensional feature space, a multidimensional feature extraction model built on deep learning algorithms is used to extract real-time merged multidimensional features of preprocessed real-time multi-source oil data.

[0013] An oil data analysis model based on deep learning algorithms is used to analyze real-time merged multidimensional features to obtain real-time oil data analysis results.

[0014] Based on the analysis results of real-time oil product data, a production, refining, storage and transportation strategy generation model based on reinforcement learning algorithm is used to generate production, refining, storage and transportation strategies, and obtain real-time production, refining, storage and transportation strategies.

[0015] Furthermore, the full life cycle data acquisition device includes several stages of data acquisition devices, and the engineering stage data acquisition device includes mining end data acquisition device, refining end data acquisition device and storage and transportation end data acquisition device.

[0016] The real-time multi-source oil data of the project includes several real-time single-source oil data. Each of the several real-time single-source oil data corresponds to a certain number of data acquisition devices at a certain stage. The real-time single-source oil data includes real-time extraction process data, real-time refining process data, and real-time inventory and logistics data.

[0017] Furthermore, using a full lifecycle data acquisition device, real-time multi-source oil product data from oil wells is collected, and the real-time multi-source oil product data is transmitted to an edge computing gateway within the communication range, including the following steps:

[0018] Using data acquisition devices at each stage of the full lifecycle data acquisition system, real-time single-source oil data from oil wells is collected and transmitted to the edge computing gateway within the communication range.

[0019] According to the OPC UA protocol, real-time multi-source oil product data, consisting of several real-time single-source oil product data, is written to the OPC UA instance in the OPC UA server.

[0020] Furthermore, using an edge computing gateway, real-time multi-source oil product data is preprocessed, and the preprocessed real-time multi-source oil product data is uploaded to the cloud data center, including the following steps:

[0021] Using an edge computing gateway, real-time multi-source oil data is sequentially cleaned, spatiotemporally aligned, and converted to obtain preprocessed real-time multi-source oil data.

[0022] Pre-processed real-time multi-source oil data is uploaded to the cloud data center via an encrypted secure channel.

[0023] Furthermore, the dynamic multidimensional feature space includes physical feature engineering, cost feature engineering, coupling feature engineering, and causal feature engineering;

[0024] Physical characteristic engineering includes parallel spatial characteristic engineering and temporal characteristic engineering;

[0025] Cost characteristic engineering includes parallel oil product value characteristic engineering, extraction cost characteristic engineering, refining cost characteristic engineering, storage and transportation cost characteristic engineering, and transaction cost characteristic engineering;

[0026] Coupling feature engineering is physical-cost feature engineering;

[0027] Causal feature engineering is causal chain-causal structure feature engineering;

[0028] The multidimensional feature extraction model is constructed based on the MPFEFN algorithm and includes a physical feature extraction module, a cost feature extraction module, a coupled feature extraction module, a causal feature extraction module, and a multidimensional feature merging module. The physical feature extraction module, cost feature extraction module, coupled feature extraction module, and causal feature extraction module correspond to the physical feature engineering, cost feature engineering, coupled feature engineering, and causal feature engineering in the dynamic multidimensional feature space, respectively. The physical feature extraction module, cost feature extraction module, coupled feature extraction module, and causal feature extraction module are all connected to the multidimensional feature merging module, which is equipped with an attention mechanism.

[0029] Furthermore, the oil product data analysis model is constructed based on the RF-MLP-MOSGA algorithm, and the oil product data analysis model includes a key feature screening module based on the RF algorithm, an oil product data analysis module based on the MLP algorithm, and an analysis result optimization module based on the MOSGA algorithm, which are connected in sequence.

[0030] Furthermore, the mining, refining, storage and transportation strategy generation model is constructed based on the MPO-MOGRPO algorithm, and the mining, refining, storage and transportation strategy generation model includes a meta-policy optimization module based on the MPO algorithm and a mining, refining, storage and transportation strategy generation module based on the MOGRPO algorithm connected in sequence. The engineering mining, refining, storage and transportation strategy generation module is equipped with a set of objective functions, an agent, a policy network and an experience playback pool.

[0031] Furthermore, in the cloud data center, based on the dynamic multidimensional feature space, a multidimensional feature extraction model built using deep learning algorithms is used to extract real-time merged multidimensional features from preprocessed real-time multi-source oil data, including the following steps:

[0032] Based on the dynamic multidimensional feature space, the preprocessed real-time multi-source oil data is input into a multidimensional feature extraction model constructed based on deep learning algorithms;

[0033] The physical feature extraction module, cost feature extraction module and coupling feature extraction module of the multidimensional feature extraction model are used to extract the real-time physical features, real-time cost features and real-time coupling features of the preprocessed real-time multi-source oil data.

[0034] Based on real-time physical characteristics, real-time cost characteristics, and real-time coupling characteristics, a causal discovery algorithm is used to identify causal chains and obtain real-time causal chains.

[0035] Based on the causal model, the causal feature extraction module of the multidimensional feature extraction model is used to extract the real-time causal features corresponding to the real-time causal chain.

[0036] Based on the dynamic attention weight values, the multidimensional feature merging module of the multidimensional feature extraction model is used to merge real-time physical features, real-time cost features, real-time coupling features, and real-time causal features to obtain real-time merged multidimensional features.

[0037] Furthermore, an oil data analysis model based on deep learning algorithms is used to analyze the real-time merged multidimensional features to obtain real-time oil data analysis results, including the following steps:

[0038] The multi-dimensional features are merged in real time and input into the oil data analysis model built based on deep learning algorithms;

[0039] The key feature filtering module of the oil data analysis model is used to extract real-time key features that are merged with multi-dimensional features in real time.

[0040] Based on real-time key features, the oil data analysis module of the oil data analysis model is used to perform analysis and prediction to obtain the real-time oil data analysis probability distribution.

[0041] The analysis result optimization module of the oil data analysis model is used to optimize the probability distribution of real-time oil data analysis and obtain the real-time oil data analysis results.

[0042] A deep learning-based oil data analysis system is provided to implement oil data analysis methods. The system includes a cloud data center, several edge computing gateways, and several full lifecycle data acquisition devices. The engineering cloud data center is connected to several edge computing gateways, and each engineering edge computing gateway is communicatively connected to a full lifecycle data acquisition device within its communication range.

[0043] The beneficial effects of this invention are as follows:

[0044] This invention provides a deep learning-based oil data analysis method and system. Through preprocessing at an edge computing gateway and a multi-dimensional feature extraction model (including physical, cost, coupling, and causal feature engineering) in a cloud data center, it can systematically and structurally integrate various data from oil wells, refineries, storage and transportation, etc., achieving data fusion and feature extraction from multi-source heterogeneous data. This avoids information silos and provides comprehensive, high-quality input for subsequent analysis. The "causal feature engineering" and "coupling feature engineering" in the dynamic multi-dimensional feature space are specifically designed to capture the interactions and causal relationships between data, making the analysis results not only based on correlation but also reflecting the inherent causal connections. This improves the interpretability of the model and the depth of understanding of the system's dynamic behavior, enhancing the systematicness, automation, and intelligence of oil data analysis. The multi-dimensional feature extraction model and oil data analysis model constructed using deep learning algorithms can automatically learn complex nonlinear relationships and deep patterns in the data, extracting more representative and predictive key features, thereby significantly improving the accuracy and robustness of analysis and prediction, especially when dealing with complex operating conditions and abnormal situations. By accelerating the initial processing of edge computing and optimizing the real-time feature extraction, analysis and optimization process in the cloud center, it is possible to respond quickly to real-time data changes in the oil production, storage and transportation process, generate real-time oil data analysis results and oil extraction, refining, storage and transportation strategies, support dynamic adjustment and instant decision-making, improve the system's flexibility and adaptability, and realize a response mechanism.

[0045] Other beneficial effects of the present invention will be further explained in the specific embodiments. Attached Figure Description

[0046] Figure 1 This is a flowchart of the deep learning-based oil data analysis method in this invention.

[0047] Figure 2 This is a structural block diagram of the deep learning-based oil data analysis system of this invention. Detailed Implementation

[0048] The present invention will be further explained below with reference to the accompanying drawings and specific embodiments.

[0049] Example 1

[0050] like Figure 1 As shown in the figure, this embodiment provides a deep learning-based oil data analysis method, including the following steps:

[0051] S1: Use a full lifecycle data acquisition device to collect real-time multi-source oil data from oil wells and transmit the real-time multi-source oil data to the edge computing gateway within the communication range;

[0052] The full life cycle data acquisition device includes several stages of data acquisition devices. The engineering stage data acquisition device includes mining end data acquisition device, refining end data acquisition device and storage and transportation end data acquisition device.

[0053] The real-time multi-source oil data of the project includes several real-time single-source oil data. Each of the several real-time single-source oil data corresponds to a certain number of data acquisition devices at a certain stage. The real-time single-source oil data includes real-time extraction process data, real-time refining process data, and real-time inventory and logistics data.

[0054] Real-time mining process data includes real-time wellhead dynamic data (three-phase motor extreme current, power harmonic distortion rate, casing pressure, tubing pressure), real-time downhole process data (bottom hole flowing pressure, pump filling rate, dynamic fluid level, sand content), and real-time crude oil physical property data (density, viscosity, water content, pour point).

[0055] Real-time refining process data includes real-time process control data (distillation tower temperature profile, fractionation tower pressure, catalyst bed temperature), real-time oil product detection data (distillation range distribution, density, flash point, sulfur content), and real-time online analysis data (near-infrared spectroscopy, mass spectrometry characteristic peaks, Fourier transform infrared spectroscopy).

[0056] Real-time inventory logistics data includes real-time tank monitoring data (liquid level, temperature, pressure, electrostatic potential), real-time pipeline transmission data (flow rate, magnetostrictive signal, residual stress), real-time environmental interaction data (temperature and humidity, illuminance, seismic wave spectrum), and real-time terminal sales data (fuel nozzle flow rate, sales transaction data, oil and gas recovery efficiency).

[0057] Using a full lifecycle data acquisition device, real-time multi-source oil product data from oil wells is collected, and the real-time multi-source oil product data is transmitted to an edge computing gateway within the communication range, including the following steps:

[0058] S1-1: Use the data acquisition devices at each stage of the full lifecycle data acquisition system to collect real-time single-source oil data from oil wells and transmit it to the edge computing gateway within the communication range.

[0059] S1-2: According to the Open Platform Communications Unified Architecture (OPC UA) protocol, real-time multi-source oil product data, consisting of several real-time single-source oil product data, is written to the OPC UA instance in the OPC UA server, including the following steps:

[0060] S1-2-1: Construct the corresponding OPC UA information model based on the physical information of all stages of the data acquisition device in the whole life cycle data acquisition device;

[0061] S1-2-2: Deploy an OPC UA server in the edge computing gateway, and create a corresponding OPC UA information model instance in the address space of the OPC UA server according to the OPC UA information model;

[0062] S1-2-3: According to the OPC UA protocol, real-time multi-source oil product data consisting of several real-time single-source oil product data is written to the OPC UA instance in the OPC UA server;

[0063] S2: Using an edge computing gateway, preprocess real-time multi-source oil data and upload the preprocessed real-time multi-source oil data to the cloud data center, including the following steps:

[0064] S2-1: Using an edge computing gateway, real-time multi-source oil data is sequentially cleaned, spatiotemporally aligned, and converted to obtain preprocessed real-time multi-source oil data;

[0065] Data cleaning rules:

[0066] Outlier handling: The Laida criterion is used in conjunction with reservoir physical property constraints (such as porosity of 0.1-0.4).

[0067] Missing value imputation: Spatial interpolation based on reservoir geological model;

[0068] Spatiotemporal alignment methods:

[0069] Timeline: ±1ms level synchronization is achieved using Precision Time Protocol (PTP) clock synchronization (IEEE1588);

[0070] Spatial axis: Establish a three-dimensional geological grid in the WGS84 coordinate system (resolution 50m×50m×10m).

[0071] Protocol conversion standards:

[0072] Edge computing gateway: Supports the OPC UA industrial protocol and converts the OPC UA protocol into a format that cloud data centers can recognize;

[0073] Cloud data centers: uniformly converted to JSON Schema v4.0 format;

[0074] S2-2: Upload pre-processed real-time multi-source oil data to the cloud data center through an encrypted secure channel;

[0075] S3: In the cloud data center, based on the dynamic multidimensional feature space, a multidimensional feature extraction model built on deep learning algorithms is used to extract real-time merged multidimensional features of preprocessed real-time multi-source oil data.

[0076] Dynamic multidimensional feature space includes physical feature engineering, cost feature engineering, coupling feature engineering, and causal feature engineering;

[0077] Physical characteristic engineering includes parallel spatial characteristic engineering and temporal characteristic engineering;

[0078] Cost characteristic engineering includes parallel oil product value characteristic engineering, extraction cost characteristic engineering, refining cost characteristic engineering, storage and transportation cost characteristic engineering, and transaction cost characteristic engineering;

[0079] Coupling feature engineering is physical-cost feature engineering;

[0080] Causal feature engineering is causal chain-causal structure feature engineering;

[0081] The multidimensional feature extraction model is built based on the Multi-scale Parallel Feature Extraction Network (MPFEFN) algorithm. The multidimensional feature extraction model includes a physical feature extraction module, a cost feature extraction module, a coupled feature extraction module, a causal feature extraction module, and a multidimensional feature merging module. The physical feature extraction module, cost feature extraction module, coupled feature extraction module, and causal feature extraction module correspond to physical feature engineering, cost feature engineering, coupled feature engineering, and causal feature engineering in the dynamic multidimensional feature space, respectively. The physical feature extraction module, cost feature extraction module, coupled feature extraction module, and causal feature extraction module are all connected to the multidimensional feature merging module. The multidimensional feature merging module is equipped with an attention mechanism.

[0082] In the cloud data center, based on the dynamic multidimensional feature space, a multidimensional feature extraction model built on deep learning algorithms is used to extract real-time merged multidimensional features from preprocessed real-time multi-source oil data, including the following steps:

[0083] S3-1: Based on the dynamic multidimensional feature space, preprocessed real-time multi-source oil data Input a multidimensional feature extraction model built based on a deep learning algorithm, where, Where N is the time indicator, M is the total number of data points, and M is the feature dimension.

[0084] S3-2: Using the physical feature extraction module, cost feature extraction module, and coupling feature extraction module of the multi-dimensional feature extraction model, real-time physical features, real-time cost features, and real-time coupling features of preprocessed real-time multi-source oil data are extracted.

[0085] The formula is:

[0086] ;

[0087] In the formula, These include real-time physical characteristics, real-time cost characteristics, and real-time coupling characteristics. These are spatial feature extraction functions and temporal feature extraction functions; These are the oil product value feature extraction functions, extraction cost feature extraction functions, refining cost feature extraction functions, storage and transportation cost feature extraction functions, and transaction cost feature extraction functions. This is the coupled feature extraction function;

[0088] S3-3: Based on real-time physical characteristics, real-time cost characteristics, and real-time coupling characteristics, a causal discovery algorithm is used to identify causal chains and obtain real-time causal chains.

[0089] The formula is:

[0090] ;

[0091] In the formula, For real-time causal chains; This is the causal discovery algorithm function;

[0092] S3-4: Based on the causal model, the causal feature extraction module of the multidimensional feature extraction model is used to extract the real-time causal features corresponding to the real-time causal chain;

[0093] The formula is:

[0094] ;

[0095] In the formula, Real-time causal characteristics; This is a causal model function, which may be a graph neural network or other model capable of processing graph structure inputs;

[0096] S3-5: Based on the dynamic attention weight values, the multidimensional feature merging module of the multidimensional feature extraction model is used to merge real-time physical features, real-time cost features, real-time coupling features, and real-time causal features to obtain real-time merged multidimensional features, including the following steps:

[0097] S3-5-1: Perform feature projection on real-time physical features, real-time cost features, real-time coupling features and real-time causal features to obtain projected real-time physical features, projected real-time cost features, projected real-time coupling features and projected real-time causal features, which are used to project features of different dimensions to a unified dimension.

[0098] The formula is:

[0099] ;

[0100] In the formula, The real-time physical features, real-time cost features, real-time coupling features, and real-time causal features after projection are defined. Here, K is the feature projection function; K is the uniform dimension.

[0101] S3-5-2: Based on the dynamic attention weight values, the multidimensional feature merging module of the multidimensional feature extraction model is used to merge the real-time physical features, real-time cost features, real-time coupling features, and real-time causal features after projection to obtain real-time merged multidimensional features.

[0102] The formula is:

[0103] ;

[0104] In the formula, To merge multidimensional features in real time; These are dynamic attention weight values;

[0105] S4: Using an oil data analysis model built on a deep learning algorithm, analyze the real-time merged multidimensional features to obtain real-time oil data analysis results;

[0106] The oil product data analysis model is built based on the Random Forest (RF)-Multi-Layer Perceptron (MLP)-Multi-Objective Snow Geese Algorithm (MOSGA) algorithm. The oil product data analysis model includes a key feature screening module built based on the RF algorithm, an oil product data analysis module built based on the MLP algorithm, and an analysis result optimization module built based on the MOSGA algorithm, which are connected in sequence.

[0107] The key feature selection module, based on calculated feature importance scores, selects the subset of features that are most critical and informative for the oil data analysis task, eliminating redundant or noisy features, thus reducing the number of input features and lowering the computational complexity and training time for subsequent analysis. The oil data analysis module can learn complex nonlinear relationships between features, used for analysis and prediction based on key features to obtain probability distributions. The analysis result optimization module is typically used to handle multi-objective optimization problems, i.e., finding an optimal set of compromise solutions among multiple conflicting objectives. In the context of oil data analysis, these objectives may include: accuracy, computational cost, error value, probability dispersion, and result complexity, etc. Iteratively improving the quality of the solution and finding the optimal result that performs well on multiple objectives is used to optimize the probability distribution of oil data analysis and improve the accuracy of oil data analysis.

[0108] An oil product data analysis model based on deep learning algorithms is used to analyze real-time merged multidimensional features to obtain real-time oil product data analysis results, including the following steps:

[0109] S4-1: Input multi-dimensional features into an oil data analysis model built on a deep learning algorithm in real time;

[0110] S4-2: Use the key feature filtering module of the oil data analysis model to extract real-time key features that are merged with multi-dimensional features in real time;

[0111] S4-3: Based on real-time key features, use the oil data analysis module of the oil data analysis model to perform analysis and prediction, and obtain the real-time oil data analysis probability distribution;

[0112] S4-4: Using the analysis result optimization module of the oil product data analysis model, optimize the probability distribution of real-time oil product data analysis to obtain the real-time oil product data analysis results, including the following steps:

[0113] S4-4-1: Analyze the probability distribution based on real-time oil data, set the format of the real-time probability distribution compensation value, and encode the initial real-time probability distribution compensation value as the individual vector of the MOSGA individual in the analysis result optimization module;

[0114] S4-4-2: Based on the individual vectors, the initialization module of the model is generated using a dynamic optimization scheme to generate several initial solutions; the initial solutions correspond to an initial real-time probability distribution compensation value.

[0115] The formula is:

[0116] ;

[0117] In the formula, The initial MOSGA individuals generated for the Circle chaotic mapping sequence, i.e., the initial solutions; For each MOSGA individual, i is a randomly generated MOSGA individual; For the remainder function;

[0118] S4-4-3: Based on the fitness function, the iterative optimization module of the model is generated using a dynamic optimization scheme. Iterative optimization is performed on several initial solutions to obtain the optimal solution, including the following steps:

[0119] S4-4-3-1: Use the fitness function to obtain the initial fitness value of each initial MOSGA individual in the initial MOSGA population, and take the initial MOSGA individual with the lowest fitness value as the leader goose.

[0120] The formula is:

[0121] ;

[0122] In the formula, The fitness function; To calculate the cost function; It is an error value function; It is the probability dispersion function; The result complexity function; It is a MOSGA individual; These are the first weight value, the second weight value, the third weight value, and the fourth weight value.

[0123] S4-4-3-2: Entering the exploration phase, a leader goose rotation mechanism, a call guidance mechanism, and a dynamic reverse mechanism are introduced to iteratively update the initial MOSGA population, resulting in an updated MOSGA population, while retaining the best individuals.

[0124] The leader goose rotation mechanism selects a new leader goose in each iteration based on the fitness values ​​of individual MOSGA individuals. This mechanism can prevent the leader goose from getting trapped in local optima too early and enhance the global search capability of the algorithm.

[0125] The formula is:

[0126] ;

[0127] In the formula, As the leading goose in an update; For the first The initial MOSGA individual with the third-to-last fitness value in the initial MOSGA population after the first iteration; The initial MOSGA individual with the fifth-to-last fitness value in the initial MOSGA population for the iteration number; This represents the current iteration number; The optimal individual; It is the first weighting factor; This is a function for generating random numbers;

[0128] The call guidance mechanism adjusts the individual position update using a sound wave propagation attenuation model based on the distance between the MOSGA individual and the leader goose. MOSGA individuals that are closer to the leader goose have a greater influence on their position update and can quickly move closer to the optimal solution, while MOSGA individuals that are farther away have a smaller influence on their position update and can maintain a certain level of exploration ability. This mechanism can avoid excessive aggregation or dispersion of the group and improve the local search accuracy of the algorithm.

[0129] The formula is:

[0130] ;

[0131] In the formula, For a newly updated MOSGA individual; For the first The initial MOSGA individuals for the number of iterations; The initial sound intensity received by the MOSGA individual; For sound intensity parameters; The initial sound intensity; The lowest acceptable sound intensity; The convergence factor; The initial MOSGA individual that is furthest away; The parameter is random. Let Brownian motion function be used. These are Brownian motion parameters; For XOR processing;

[0132] ;

[0133] In the formula, is the convergence factor; tanh(.) is the hyperbolic tangent function; This represents the current iteration number; a is the maximum number of iterations. max a min λ represents the maximum and minimum values ​​of the convergence factor, respectively; λ is the deceleration rate parameter. As a decreasing periodic parameter, λ = -2π. ;

[0134] The dynamic reverse mechanism dynamically reverses the initial MOSGA individuals, increasing the diversity of exploration directions and avoiding getting trapped in local optima;

[0135] The formula is:

[0136] ;

[0137] In the formula, For a single update of the reverse MOSGA individual; γ is the decreasing inertia coefficient; L max L min These are the maximum and minimum values ​​in the vector space, respectively.

[0138] The leader goose from the first update, several MOSGA individuals from the first update, and several reverse MOSGA individuals from the first update will be integrated to obtain a MOSGA population from the first update, and the MOSGA individual with the lowest fitness value will be retained as the best individual.

[0139] S4-4-3-3: Entering the development phase, anomaly boundary strategy and Gaussian mutation mechanism are introduced to perform a second update on the MOSGA population updated once, resulting in a second-updated MOSGA population, and the best individual is retained.

[0140] The abnormal boundary strategy calculates the difference between the fitness value of each updated MOSGA individual and the population average fitness value. For MOSGA individuals with fitness values ​​much higher than the population average, their position update method will be adjusted, such as using Gaussian mutation mechanism, larger step size or smaller step size. This mechanism can help individuals avoid getting trapped in local optima and improve the convergence speed and accuracy of the algorithm.

[0141] The formula is:

[0142] ;

[0143] In the formula, This is a MOSGA individual that has undergone a second update; For a newly updated MOSGA individual; The fitness function; This represents the average fitness value of the population. The individual with the highest fitness value is the MOSGA. These are the second and third weighting factors; These are parameters for the Gaussian mutation mechanism;

[0144] S4-4-3-4: If the number of iterations is greater than or equal to the iteration number threshold or the fitness value of the best individual is less than the fitness threshold, then the best individual will be output as the optimal solution.

[0145] S4-4-4: Decode the individual vector of the optimal solution to obtain the optimal real-time probability distribution compensation value, and optimize the probability distribution of real-time oil data analysis based on the optimal real-time probability distribution compensation value to obtain the real-time oil data analysis result.

[0146] S5: Based on the analysis results of real-time oil product data, a production, refining, storage and transportation strategy generation model based on reinforcement learning algorithm is used to generate production, refining, storage and transportation strategies, and obtain real-time production, refining, storage and transportation strategies.

[0147] The mining, refining, storage and transportation strategy generation model is constructed based on the Meta-Policy Optimization (MPO) - Multi-Objective Group Relative Policy Optimization (MOGRPO) algorithm. The mining, refining, storage and transportation strategy generation model includes a meta-policy optimization module based on the MPO algorithm and a mining, refining, storage and transportation strategy generation module based on the MOGRPO algorithm, which are connected in sequence. The engineering mining, refining, storage and transportation strategy generation module is equipped with a set of objective functions, an agent, a policy network and an experience replay pool.

[0148] The meta-policy optimization module optimizes the network parameters of the policy network in the mining, refining, storage, and transportation policy generation module. This allows these parameters to quickly adapt to new and unseen performance detection prediction results, improving the model's generalization ability. Even under unseen performance detection prediction results, the policy network can be updated based on previous learning experience, enhancing the adaptability of the mining, refining, storage, and transportation policy generation model. The objective function set of the mining, refining, storage, and transportation policy generation module can handle multiple conflicting objectives, such as response time, storage and transportation costs, and transaction impact, generating mining, refining, storage, and transportation policies that balance these objectives. The agent learns from historical mining, refining, storage, and transportation policies through an experience replay pool, continuously optimizing its own policy generation capabilities. The agent controls the policy network based on the learned experience to generate more effective mining, refining, storage, and transportation policies. The design of the experience replay pool and the agent enables the model to continuously learn and optimize, improving the quality of policy generation. Because the mining, refining, storage, and transportation policy generation module adopts a group exploration approach, it can avoid getting trapped in local optima to some extent. The policy network outputs the probability distribution of actions in a given state. The mining, refining, storage, and transportation policy generation module directly updates the policy network through gradients, eliminating the value network in traditional reinforcement learning, making the algorithm structure simpler.

[0149] Based on the analysis results of real-time oil product data, a production, refining, storage and transportation strategy generation model based on reinforcement learning algorithm is used to generate production, refining, storage and transportation strategies, resulting in real-time production, refining, storage and transportation strategies. The process includes the following steps:

[0150] S5-1: Based on the real-time oil product data analysis results, the meta-strategy optimization module of the oil extraction, refining, storage and transportation strategy generation model is used to update the strategy network of the oil extraction, refining, storage and transportation strategy generation module to obtain the updated strategy network.

[0151] S5-2: Randomly extract several historical mining, refining, storage and transportation strategy generation experiences from the experience replay pool, generate several possible mining, refining, storage and transportation decision actions based on these historical mining, refining, storage and transportation strategy generation experiences, and update the action space of the intelligent agent of the mining, refining, storage and transportation strategy generation module based on these possible mining, refining, storage and transportation decision actions to obtain the updated action space.

[0152] S5-3: Analyze the real-time oil data analysis results to obtain several real-time oil data states, and update the state space of the intelligent agent of the oil extraction, refining, storage and transportation strategy generation module according to the real-time oil data states to obtain the updated state space.

[0153] S5-4: Select a real-time objective function from the objective function set of the mining, refining, storage and transportation strategy generation module, and based on the real-time objective function, use the updated agent of the mining, refining, storage and transportation strategy generation module to control the updated strategy network, and generate the probability distribution of all possible mining, refining, storage and transportation decision actions in the updated action space corresponding to each real-time oil product data state in the updated state space.

[0154] S5-5: Take the most probable mining, refining, storage and transportation decision action with the highest probability distribution in the updated action space as the corresponding real-time oil product data state to execute the mining, refining, storage and transportation decision action. Integrate all the executed mining, refining, storage and transportation decision actions of the real-time oil product data state in the updated state space to obtain the real-time mining, refining, storage and transportation strategy.

[0155] Example 2

[0156] like Figure 2 As shown, this embodiment provides a deep learning-based oil data analysis system for implementing oil data analysis methods. The system includes a cloud data center, several edge computing gateways, and several full lifecycle data acquisition devices. The engineering cloud data center is connected to several edge computing gateways, and each engineering edge computing gateway is communicatively connected to a full lifecycle data acquisition device within its communication range.

[0157] The full lifecycle data acquisition device is used to collect real-time multi-source oil data from oil wells and transmit the real-time multi-source oil data to the edge computing gateway within the communication range;

[0158] Edge computing gateways are used to preprocess real-time multi-source oil data and upload the preprocessed real-time multi-source oil data to the cloud data center.

[0159] The cloud data center is used to extract real-time merged multidimensional features from preprocessed real-time multi-source oil data using a multidimensional feature extraction model built on a deep learning algorithm, based on a dynamic multidimensional feature space; to analyze the real-time merged multidimensional features using an oil data analysis model built on a deep learning algorithm, and to obtain real-time oil data analysis results; based on the real-time oil data analysis results, to generate mining, refining, storage and transportation strategies using a mining, refining, storage and transportation strategy generation model built on a reinforcement learning algorithm, and to obtain real-time mining, refining, storage and transportation strategies.

[0160] This invention provides a deep learning-based oil data analysis method and system. Through preprocessing at an edge computing gateway and a multi-dimensional feature extraction model (including physical, cost, coupling, and causal feature engineering) in a cloud data center, it can systematically and structurally integrate various data from oil wells, refineries, storage and transportation, etc., achieving data fusion and feature extraction from multi-source heterogeneous data. This avoids information silos and provides comprehensive, high-quality input for subsequent analysis. The "causal feature engineering" and "coupling feature engineering" in the dynamic multi-dimensional feature space are specifically designed to capture the interactions and causal relationships between data, making the analysis results not only based on correlation but also reflecting the inherent causal connections. This improves the interpretability of the model and the depth of understanding of the system's dynamic behavior, enhancing the systematicness, automation, and intelligence of oil data analysis. The multi-dimensional feature extraction model and oil data analysis model constructed using deep learning algorithms can automatically learn complex nonlinear relationships and deep patterns in the data, extracting more representative and predictive key features, thereby significantly improving the accuracy and robustness of analysis and prediction, especially when dealing with complex operating conditions and abnormal situations. By accelerating the initial processing of edge computing and optimizing the real-time feature extraction, analysis and optimization process in the cloud center, it is possible to respond quickly to real-time data changes in the oil production, storage and transportation process, generate real-time oil data analysis results and oil extraction, refining, storage and transportation strategies, support dynamic adjustment and instant decision-making, improve the system's flexibility and adaptability, and realize a response mechanism.

[0161] This invention is not limited to the optional embodiments described above, and anyone can derive other various forms of products based on the inspiration of this invention. The specific embodiments described above should not be construed as limiting the scope of protection of this invention; the scope of protection of this invention should be determined by the claims, and the specification can be used to interpret the claims.

Claims

1. A deep learning-based method for oil product data analysis, characterized in that: Includes the following steps: Using a full lifecycle data acquisition device, real-time multi-source oil data from oil wells is collected and transmitted to an edge computing gateway within the communication range. Using an edge computing gateway, real-time multi-source oil data is preprocessed, and the preprocessed real-time multi-source oil data is uploaded to the cloud data center. In the cloud data center, based on the dynamic multidimensional feature space, a multidimensional feature extraction model built on deep learning algorithms is used to extract real-time merged multidimensional features of preprocessed real-time multi-source oil data. The dynamic multidimensional feature space includes physical feature engineering, cost feature engineering, coupling feature engineering, and causal feature engineering; The physical characteristic engineering includes parallel spatial characteristic engineering and temporal characteristic engineering; The cost characteristic engineering includes parallel oil product value characteristic engineering, extraction cost characteristic engineering, refining cost characteristic engineering, storage and transportation cost characteristic engineering, and transaction cost characteristic engineering; The coupling feature engineering mentioned above is physical-cost feature engineering; The aforementioned causal feature engineering is causal chain-causal structure feature engineering; The multidimensional feature extraction model is constructed based on the MPFEFN algorithm and includes a physical feature extraction module, a cost feature extraction module, a coupling feature extraction module, a causal feature extraction module, and a multidimensional feature merging module. The physical feature extraction module, cost feature extraction module, coupling feature extraction module, and causal feature extraction module correspond to the physical feature engineering, cost feature engineering, coupling feature engineering, and causal feature engineering in the dynamic multidimensional feature space, respectively. The physical feature extraction module, cost feature extraction module, coupling feature extraction module, and causal feature extraction module are all connected to the multidimensional feature merging module, which is equipped with an attention mechanism. Includes the following steps: Based on the dynamic multidimensional feature space, the preprocessed real-time multi-source oil data is input into a multidimensional feature extraction model constructed based on deep learning algorithms; The physical feature extraction module, cost feature extraction module and coupling feature extraction module of the multidimensional feature extraction model are used to extract the real-time physical features, real-time cost features and real-time coupling features of the preprocessed real-time multi-source oil data. Based on real-time physical characteristics, real-time cost characteristics, and real-time coupling characteristics, a causal discovery algorithm is used to identify causal chains and obtain real-time causal chains. Based on the causal model, the causal feature extraction module of the multidimensional feature extraction model is used to extract the real-time causal features corresponding to the real-time causal chain. Based on the dynamic attention weight values, the multidimensional feature merging module of the multidimensional feature extraction model is used to merge real-time physical features, real-time cost features, real-time coupling features, and real-time causal features to obtain real-time merged multidimensional features. An oil data analysis model based on deep learning algorithms is used to analyze real-time merged multidimensional features to obtain real-time oil data analysis results. Based on the analysis results of real-time oil product data, a production, refining, storage and transportation strategy generation model based on reinforcement learning algorithm is used to generate production, refining, storage and transportation strategies, and obtain real-time production, refining, storage and transportation strategies.

2. The oil data analysis method based on deep learning according to claim 1, characterized in that: The full life cycle data acquisition device includes several stage data acquisition devices, including mining end data acquisition devices, refining end data acquisition devices, and storage and transportation end data acquisition devices. The real-time multi-source oil data includes several real-time single-source oil data, each of which corresponds to several stages of data acquisition devices. The real-time single-source oil data includes real-time extraction process data, real-time refining process data, and real-time inventory and logistics data.

3. The oil data analysis method based on deep learning according to claim 2, characterized in that: Using a full lifecycle data acquisition device, real-time multi-source oil product data from oil wells is collected, and the real-time multi-source oil product data is transmitted to an edge computing gateway within the communication range, including the following steps: Using data acquisition devices at each stage of the full lifecycle data acquisition system, real-time single-source oil data from oil wells is collected and transmitted to the edge computing gateway within the communication range. According to the OPC UA protocol, real-time multi-source oil product data, consisting of several real-time single-source oil product data, is written to the OPC UA instance in the OPC UA server.

4. The oil data analysis method based on deep learning according to claim 3, characterized in that: Using an edge computing gateway, real-time multi-source oil data is preprocessed, and the preprocessed real-time multi-source oil data is uploaded to the cloud data center, including the following steps: Using an edge computing gateway, real-time multi-source oil data is sequentially cleaned, spatiotemporally aligned, and converted to obtain preprocessed real-time multi-source oil data. Pre-processed real-time multi-source oil data is uploaded to the cloud data center via an encrypted secure channel.

5. The oil data analysis method based on deep learning according to claim 4, characterized in that: The oil product data analysis model is constructed based on the RF-MLP-MOSGA algorithm, and includes a key feature screening module based on the RF algorithm, an oil product data analysis module based on the MLP algorithm, and an analysis result optimization module based on the MOSGA algorithm, which are connected in sequence.

6. The oil data analysis method based on deep learning according to claim 5, characterized in that: The mining, refining, storage and transportation strategy generation model is constructed based on the MPO-MOGRPO algorithm, and the mining, refining, storage and transportation strategy generation model includes a meta-policy optimization module based on the MPO algorithm and a mining, refining, storage and transportation strategy generation module based on the MOGRPO algorithm connected in sequence. The mining, refining, storage and transportation strategy generation module is equipped with a set of objective functions, an agent, a policy network and an experience replay pool.

7. The oil data analysis method based on deep learning according to claim 6, characterized in that: An oil product data analysis model based on deep learning algorithms is used to analyze real-time merged multidimensional features to obtain real-time oil product data analysis results, including the following steps: The multi-dimensional features are merged in real time and input into the oil data analysis model built based on deep learning algorithms; The key feature filtering module of the oil data analysis model is used to extract real-time key features that are merged with multi-dimensional features in real time. Based on real-time key features, the oil data analysis module of the oil data analysis model is used to perform analysis and prediction to obtain the real-time oil data analysis probability distribution. The analysis result optimization module of the oil data analysis model is used to optimize the probability distribution of real-time oil data analysis and obtain the real-time oil data analysis results.

8. A deep learning-based oil data analysis system for implementing the oil data analysis method as described in any one of claims 1-7, characterized in that: The system includes a cloud data center, several edge computing gateways, and several full lifecycle data acquisition devices. The cloud data center is connected to several edge computing gateways, and each edge computing gateway is communicatively connected to a full lifecycle data acquisition device within its communication range.

Images